[philippose]

Hello Ning,

It could mean that your flow is actually not laminar, but turbulent.... due to which the laminar solver is not able to converge.

Your mesh seems to be quite ok....

Maybe you can try a turbulent simulation using simpleFoam and see if that converges.

Enjoy!

Philippose

[nzy102]

Hi Philippose:

Thank you for your input. The Re number for my case is only around 1000. I don't see any possibility of turbulent flows.

Ning

[philippose]

Hello again Ning,

Since I dont know anything about the case you are trying to simulate, I need to be cautious about this....

I think a Reynolds Number of 1000 is in a region where you need to start worrying about the flow already being turbulent, or being somewhere in the transition region.

The typical Reynolds Number of 2300 as the limit for laminar flow that one keeps reading about, is for a very specific case, of a smooth straight pipe.

There are systems where you can have a critical Reynolds Number as low as 500....

Usually, if the Reynolds Number is "very high", you can easily say that the flow is turbulent, or if the Reynolds Number is "very low", you can call the flow laminar... but in the region that you are in..... one needs to be quite careful.

Since you specified that you are working on highly curved 3D flow... with more than one input and more than one output (implying that there will be flow mixing, and dividing...etc....etc...), I think you need to consider the possibility of turbulence in your system.

Have a nice day!

Philippose

[nzy102]

Thanks philippose. I think the flow in my case is more like in a transitional state. It is not possible to use the current available low-Re two equation or other turbulent models to study this type of flow. There is probably some unsteadiness with the flow. But the problem is that I got good converged results using first-order upwind for steady laminar flow. Somehow it doesn't converge for the second order schemes. Anybody here can give me some hints? I am really stuck with this problem for a while. Thank you.

Ning

[nzy102]

Hi Philippose:

I tried some low-Re turbulence models (lienLescchz and liencubic). I got the same thing. It converges well for the first-oder upwind but not for second-order schemes. The geometry I am using is pretty complex, and first-order scheme apparently is not accurate enough, not sure how to deal with this case now.